INDEX

Explanations

phrases that refer to something mentioned or described previously in a text

oai_token-act-pair · gpt-3.5-turbo

references to preceding content or information

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

aceous

-0.71

ãĤ¼ãĤ¦ãĤ¹

-0.70

 Rounds

-0.64

iform

-0.63

Mail

-0.61

Stars

-0.61

 Invasion

-0.60

eness

-0.60

ately

-0.59

Introduced

-0.59

POSITIVE LOGITS

mentioned

1.29

ground

1.27

 mentioned

0.82

spection

0.73

lihood

0.72

perty

0.71

above

0.70

 âĨĳ

0.69

nder

0.68

ebin

0.66

Activations Density 0.037%

phrases that refer to something mentioned or described previously in a text

references to preceding content or information

No Comments

No Known Activations